Two-Sample Tests of Location
In Chapters 12 and 13 we presented a number of one-sample procedures for addressing questions of location, spread, and relationship. We presented the traditional, distribution-based approaches, the non-parametric approaches where available, and the newer, computer-intensive resampling approaches. In this chapter we are going to follow the same paradigm, with the additional consideration of independent versus dependent samples. Independent samples may arise in a number of different ways. First, one might draw a random sample from a population and randomly assign the cases to one of two conditions. For example, we may assign half to an experimental group and half to a control group. In another situation, one might draw a random sample from a population of females and a separate random sample from a population of males. In each of these two cases, the observations in each of the two groups would be considered to be independent.